Add new dependencies and update existing ones

app.ipynb

@@ -0,0 +1,377 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from datasets import load_dataset\n",
+    "from trl import SFTTrainer\n",
+    "from peft import LoraConfig, get_peft_model\n",
+    "\n",
+    "import os\n",
+    "from uuid import uuid4\n",
+    "import pandas as pd\n",
+    "\n",
+    "import subprocess\n",
+    "import evaluate\n",
+    "import transformers\n",
+    "from transformers import AutoModelForCausalLM, AutoTokenizer\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def max_token_len(dataset):\n",
+    "    max_seq_length = 0\n",
+    "    for row in dataset:\n",
+    "        tokens = len(tokenizer(row['text'])['input_ids'])\n",
+    "        if tokens > max_seq_length:\n",
+    "            max_seq_length = tokens\n",
+    "    return max_seq_length"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Model Max Length: 1000000000000000019884624838656\n"
+     ]
+    }
+   ],
+   "source": [
+    "# model_name='TinyLlama/TinyLlama-1.1B-Chat-v0.1'\n",
+    "model_name = 'mistralai/Mistral-7B-v0.1'\n",
+    "# model_name = 'distilbert-base-uncased'\n",
+    "tokenizer = AutoTokenizer.from_pretrained(model_name)\n",
+    "model_max_length = tokenizer.model_max_length\n",
+    "print(\"Model Max Length:\", model_max_length)\n",
+    "\n",
+    "# dataset = load_dataset(\"imdb\", split=\"train\")\n",
+    "dataset_name = 'ai-aerospace/ams_data_train_generic_v0.1_100'\n",
+    "dataset = load_dataset(dataset_name)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Max token length train: 1121\n",
+      "Max token length validation: 38\n",
+      "Block size: 2242\n",
+      "{'project_name': './llms/ams_data_train-100_4ba55532-e0b2-478b-9f5b-beb082e1b557', 'model_name': 'mistralai/Mistral-7B-v0.1', 'repo_id': 'ai-aerospace/ams-data-train-100-11b94ea4-2b2b-4db3-9e69-acb5a5d9f3e8', 'train_data': 'train_data', 'data_directory': './fine_tune_data/', 'block_size': 2242, 'model_max_length': 1121, 'logging_steps': -1, 'evaluation_strategy': 'epoch', 'save_total_limit': 1, 'save_strategy': 'epoch', 'mixed_precision': 'fp16', 'lr': 3e-05, 'epochs': 3, 'batch_size': 2, 'warmup_ratio': 0.1, 'gradient_accumulation': 1, 'optimizer': 'adamw_torch', 'scheduler': 'linear', 'weight_decay': 0, 'max_grad_norm': 1, 'seed': 42, 'quantization': 'int4', 'lora_r': 16, 'lora_alpha': 32, 'lora_dropout': 0.05}\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Write dataset files into data directory\n",
+    "data_directory = './fine_tune_data/'\n",
+    "\n",
+    "# Create the data directory if it doesn't exist\n",
+    "os.makedirs(data_directory, exist_ok=True)\n",
+    "\n",
+    "# Write the train data to a CSV file\n",
+    "train_data='train_data'\n",
+    "train_filename = os.path.join(data_directory, train_data)\n",
+    "dataset['train'].to_pandas().to_csv(train_filename+'.csv', columns=['text'], index=False)\n",
+    "max_token_length_train=max_token_len(dataset['train'])\n",
+    "print('Max token length train: '+str(max_token_length_train))\n",
+    "\n",
+    "# Write the validation data to a CSV file\n",
+    "validation_data='validation_data'\n",
+    "validation_filename = os.path.join(data_directory, validation_data)\n",
+    "dataset['validation'].to_pandas().to_csv(validation_filename+'.csv', columns=['text'], index=False)\n",
+    "max_token_length_validation=max_token_len(dataset['validation'])\n",
+    "print('Max token length validation: '+str(max_token_length_validation))\n",
+    "      \n",
+    "max_token_length=max(max_token_length_train,max_token_length_validation)\n",
+    "# max_token_length=max_token_length_train\n",
+    "if max_token_length > model_max_length:\n",
+    "    raise ValueError(\"Maximum token length exceeds model limits.\")\n",
+    "block_size=2*max_token_length\n",
+    "print('Block size: '+str(block_size))\n",
+    "\n",
+    "# Define project parameters\n",
+    "username='ai-aerospace'\n",
+    "project_name='./llms/'+'ams_data_train-100_'+str(uuid4())\n",
+    "repo_name='ams-data-train-100-'+str(uuid4())\n",
+    "\n",
+    "model_params={\n",
+    "  \"project_name\": project_name,\n",
+    "  \"model_name\": model_name,\n",
+    "  \"repo_id\": username+'/'+repo_name,\n",
+    "  \"train_data\": train_data,\n",
+    "  \"validation_data\": validation_data,\n",
+    "  \"data_directory\": data_directory,\n",
+    "  \"block_size\": block_size,\n",
+    "  \"model_max_length\": max_token_length,\n",
+    "  \"logging_steps\": -1,\n",
+    "  \"evaluation_strategy\": \"epoch\",\n",
+    "  \"save_total_limit\": 1,\n",
+    "  \"save_strategy\": \"epoch\",\n",
+    "  \"mixed_precision\": \"fp16\",\n",
+    "  \"lr\": 0.00003,\n",
+    "  \"epochs\": 3,\n",
+    "  \"batch_size\": 2,\n",
+    "  \"warmup_ratio\": 0.1,\n",
+    "  \"gradient_accumulation\": 1,\n",
+    "  \"optimizer\": \"adamw_torch\",\n",
+    "  \"scheduler\": \"linear\",\n",
+    "  \"weight_decay\": 0,\n",
+    "  \"max_grad_norm\": 1,\n",
+    "  \"seed\": 42,\n",
+    "  \"quantization\": \"int4\",\n",
+    "  \"lora_r\": 16,\n",
+    "  \"lora_alpha\": 32,\n",
+    "  \"lora_dropout\": 0.05\n",
+    "}\n",
+    "for key, value in model_params.items():\n",
+    "  os.environ[key] = str(value)\n",
+    "\n",
+    "print(model_params)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "### Start trainer\n",
+    "# trainer = SFTTrainer(\n",
+    "#     model_name,\n",
+    "#     train_dataset=dataset,\n",
+    "#     dataset_text_field=\"text\",\n",
+    "#     max_seq_length=512,\n",
+    "# )\n",
+    "\n",
+    "peft_config = LoraConfig(\n",
+    "    r=model_params['lora_r'],\n",
+    "    lora_alpha=model_params['lora_alpha'],\n",
+    "    lora_dropout=model_params['lora_dropout']\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Load the model\n",
+    "model = AutoModelForCausalLM.from_pretrained(\n",
+    "    model_name,\n",
+    "    load_in_4bit=True\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Setting up the LoRA model\n",
+    "# import os\n",
+    "# os.environ[\"CUDA_VISIBLE_DEVICES\"]=\"0\"\n",
+    "# from transformers import AutoModelForSequenceClassification\n",
+    "# from peft import LoraConfig, get_peft_model, TaskType\n",
+    "\n",
+    "# MODEL =\"xlm-roberta-large\"\n",
+    "\n",
+    "# config = LoraConfig(\n",
+    "#     task_type=\"SEQ_CLS\",\n",
+    "#     r=16,\n",
+    "#     lora_alpha=16,\n",
+    "#     target_modules=[\"query\", \"value\"],  # Targets the attention blocks in the model\n",
+    "#     lora_dropout=0.1,\n",
+    "#     bias=\"none\",\n",
+    "#     modules_to_save=[\"classifier\"],\n",
+    "# )\n",
+    "\n",
+    "# model = AutoModelForSequenceClassification.from_pretrained(\n",
+    "#     MODEL,\n",
+    "#     num_labels=len(unique_subissues),\n",
+    "#     id2label=id2label,\n",
+    "#     label2id=label2id,\n",
+    "#     ignore_mismatched_sizes=True\n",
+    "# )  \n",
+    "\n",
+    "lora_model = get_peft_model(model, peft_config)\n",
+    "lora_model.print_trainable_parameters()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "trainer = SFTTrainer(\n",
+    "    model,\n",
+    "    train_dataset=dataset,\n",
+    "    dataset_text_field=\"text\",\n",
+    "    peft_config=peft_config,\n",
+    "    max_seq_length=model_params['model_max_length']\n",
+    ")\n",
+    "\n",
+    "trainer.train()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "4fbe714ca43d4e53aec27f4ce4fb4706",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading builder script:   0%|          | 0.00/6.77k [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "826f51589454434b891a94b0d5ef8a73",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading builder script:   0%|          | 0.00/7.36k [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "81418551f332492293ee9795f98a62f7",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading builder script:   0%|          | 0.00/4.20k [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "application/vnd.jupyter.widget-view+json": {
+       "model_id": "367f897f76f845d782ebc3f9be4eec4d",
+       "version_major": 2,
+       "version_minor": 0
+      },
+      "text/plain": [
+       "Downloading builder script:   0%|          | 0.00/7.55k [00:00<?, ?B/s]"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "ename": "NameError",
+     "evalue": "name 'lora_model' is not defined",
+     "output_type": "error",
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
+      "Cell \u001b[0;32mIn[8], line 18\u001b[0m\n\u001b[1;32m     13\u001b[0m     results\u001b[38;5;241m.\u001b[39mupdate(precision_metric\u001b[38;5;241m.\u001b[39mcompute(predictions\u001b[38;5;241m=\u001b[39mpredictions, references \u001b[38;5;241m=\u001b[39m labels, average\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mmacro\u001b[39m\u001b[38;5;124m\"\u001b[39m))\n\u001b[1;32m     15\u001b[0m     \u001b[38;5;28;01mreturn\u001b[39;00m results\n\u001b[1;32m     17\u001b[0m trainer \u001b[38;5;241m=\u001b[39m transformers\u001b[38;5;241m.\u001b[39mTrainer(\n\u001b[0;32m---> 18\u001b[0m     model\u001b[38;5;241m=\u001b[39m\u001b[43mlora_model\u001b[49m,\n\u001b[1;32m     19\u001b[0m     train_dataset\u001b[38;5;241m=\u001b[39mtrain_dataset,\n\u001b[1;32m     20\u001b[0m     eval_dataset\u001b[38;5;241m=\u001b[39mval_dataset,\n\u001b[1;32m     21\u001b[0m     compute_metrics\u001b[38;5;241m=\u001b[39mcompute_metrics,\n\u001b[1;32m     22\u001b[0m     args\u001b[38;5;241m=\u001b[39mtransformers\u001b[38;5;241m.\u001b[39mTrainingArguments(\n\u001b[1;32m     23\u001b[0m         per_device_train_batch_size\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m8\u001b[39m,\n\u001b[1;32m     24\u001b[0m         per_device_eval_batch_size\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m32\u001b[39m,\n\u001b[1;32m     25\u001b[0m         gradient_accumulation_steps\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m4\u001b[39m,\n\u001b[1;32m     26\u001b[0m         warmup_steps\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m100\u001b[39m,\n\u001b[1;32m     27\u001b[0m         max_steps\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m12276\u001b[39m,\n\u001b[1;32m     28\u001b[0m         learning_rate\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m2e-4\u001b[39m,\n\u001b[1;32m     29\u001b[0m         fp16\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m     30\u001b[0m         eval_steps\u001b[38;5;241m=\u001b[39m \u001b[38;5;241m1000\u001b[39m,\n\u001b[1;32m     31\u001b[0m         logging_steps\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m1000\u001b[39m,\n\u001b[1;32m     32\u001b[0m         save_steps\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m1000\u001b[39m,\n\u001b[1;32m     33\u001b[0m         evaluation_strategy\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124msteps\u001b[39m\u001b[38;5;124m\"\u001b[39m,\n\u001b[1;32m     34\u001b[0m         do_eval\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m     35\u001b[0m         load_best_model_at_end\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m,\n\u001b[1;32m     36\u001b[0m         metric_for_best_model\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mf1\u001b[39m\u001b[38;5;124m\"\u001b[39m,\n\u001b[1;32m     37\u001b[0m         output_dir\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mmodel_outputs\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m     38\u001b[0m         logging_dir\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mmodel_outputs\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m     39\u001b[0m         remove_unused_columns \u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mFalse\u001b[39;00m, \n\u001b[1;32m     40\u001b[0m         report_to\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mwandb\u001b[39m\u001b[38;5;124m'\u001b[39m  \u001b[38;5;66;03m# enable logging to W&B\u001b[39;00m\n\u001b[1;32m     41\u001b[0m     ),\n\u001b[1;32m     42\u001b[0m )\n\u001b[1;32m     43\u001b[0m trainer\u001b[38;5;241m.\u001b[39mtrain()\n",
+      "\u001b[0;31mNameError\u001b[0m: name 'lora_model' is not defined"
+     ]
+    }
+   ],
+   "source": [
+    "f1_metric = evaluate.load(\"f1\")\n",
+    "recall_metric = evaluate.load(\"recall\")\n",
+    "accuracy_metric = evaluate.load(\"accuracy\")\n",
+    "precision_metric = evaluate.load(\"precision\")\n",
+    "\n",
+    "def compute_metrics(eval_pred):\n",
+    "    logits, labels = eval_pred\n",
+    "    predictions = np.argmax(logits, axis=-1)\n",
+    "    results = {}\n",
+    "    results.update(f1_metric.compute(predictions=predictions, references = labels, average=\"macro\"))\n",
+    "    results.update(recall_metric.compute(predictions=predictions, references = labels, average=\"macro\"))\n",
+    "    results.update(accuracy_metric.compute(predictions=predictions, references = labels))\n",
+    "    results.update(precision_metric.compute(predictions=predictions, references = labels, average=\"macro\"))\n",
+    "\n",
+    "    return results\n",
+    "\n",
+    "# See https://towardsdatascience.com/fine-tune-your-llm-without-maxing-out-your-gpu-db2278603d78 for details\n",
+    "trainer = transformers.Trainer(\n",
+    "    model=lora_model,\n",
+    "    train_dataset=model_params['train_data'],\n",
+    "    eval_dataset=model_params['validation_data'],\n",
+    "    compute_metrics=compute_metrics,\n",
+    "    args=transformers.TrainingArguments(\n",
+    "        per_device_train_batch_size=model_params['batch_size'],\n",
+    "        per_device_eval_batch_size=model_params['batch_size'],\n",
+    "        gradient_accumulation_steps=model_params['gradient_accumulation'],\n",
+    "        warmup_steps=100,\n",
+    "        max_steps=12276,\n",
+    "        learning_rate=model_params['lr'],\n",
+    "        fp16=True,\n",
+    "        eval_steps= 1000,\n",
+    "        logging_steps=1000,\n",
+    "        save_steps=1000,\n",
+    "        evaluation_strategy=model_params['evaluation_strategy'],\n",
+    "        do_eval=True,\n",
+    "        load_best_model_at_end=True,\n",
+    "        metric_for_best_model=\"f1\",\n",
+    "        output_dir='model_outputs',\n",
+    "        logging_dir='model_outputs',\n",
+    "        remove_unused_columns =False, \n",
+    "        report_to='wandb'  # enable logging to W&B\n",
+    "    ),\n",
+    ")\n",
+    "trainer.train()"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": ".venv",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.11.7"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

app.py

@@ -1,14 +1,23 @@
 from datasets import load_dataset
 from trl import SFTTrainer
-from peft import LoraConfig
+from peft import LoraConfig, get_peft_model
 
 import os
 from uuid import uuid4
 import pandas as pd
 
 import subprocess
+import transformers
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
+import evaluate
+from sklearn.metrics import f1_score, precision_score, recall_score, accuracy_score
+
+from datasets import load_dataset
+from trl import SFTTrainer
+from peft import LoraConfig, get_peft_model
+
+### Define functions
 def max_token_len(dataset):
     max_seq_length = 0
     for row in dataset:
@@ -17,6 +26,7 @@ def max_token_len(dataset):
             max_seq_length = tokens
     return max_seq_length
 
+### Set up models and datasets, training parameters
 # model_name='TinyLlama/TinyLlama-1.1B-Chat-v0.1'
 model_name = 'mistralai/Mistral-7B-v0.1'
 # model_name = 'distilbert-base-uncased'
@@ -26,7 +36,8 @@ print("Model Max Length:", model_max_length)
 
 # dataset = load_dataset("imdb", split="train")
 dataset_name = 'ai-aerospace/ams_data_train_generic_v0.1_100'
-dataset = load_dataset(dataset_name, split="train")
+dataset = load_dataset(dataset_name)
+
 
 # Write dataset files into data directory
 data_directory = './fine_tune_data/'
@@ -49,6 +60,7 @@ max_token_length_validation=max_token_len(dataset['validation'])
 print('Max token length validation: '+str(max_token_length_validation))
       
 max_token_length=max(max_token_length_train,max_token_length_validation)
+# max_token_length=max_token_length_train
 if max_token_length > model_max_length:
     raise ValueError("Maximum token length exceeds model limits.")
 block_size=2*max_token_length
@@ -93,32 +105,61 @@ for key, value in model_params.items():
 
 print(model_params)
 
-### Load model
+### Load model and peft config, calculate trainable parameters
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
     load_in_4bit=True
 )
-
-### Start trainer
-# trainer = SFTTrainer(
-#     model_name,
-#     train_dataset=dataset,
-#     dataset_text_field="text",
-#     max_seq_length=512,
-# )
-
 peft_config = LoraConfig(
     r=model_params['lora_r'],
     lora_alpha=model_params['lora_alpha'],
     lora_dropout=model_params['lora_dropout']
 )
-
-trainer = SFTTrainer(
-    model,
-    train_dataset=dataset,
-    dataset_text_field="text",
-    peft_config=peft_config,
-    max_seq_length=model_params['model_max_length']
+lora_model = get_peft_model(model, peft_config)
+lora_model.print_trainable_parameters()
+
+### Train model
+f1_metric = evaluate.load("f1")
+recall_metric = evaluate.load("recall")
+accuracy_metric = evaluate.load("accuracy")
+precision_metric = evaluate.load("precision")
+
+def compute_metrics(eval_pred):
+    logits, labels = eval_pred
+    predictions = np.argmax(logits, axis=-1)
+    results = {}
+    results.update(f1_metric.compute(predictions=predictions, references = labels, average="macro"))
+    results.update(recall_metric.compute(predictions=predictions, references = labels, average="macro"))
+    results.update(accuracy_metric.compute(predictions=predictions, references = labels))
+    results.update(precision_metric.compute(predictions=predictions, references = labels, average="macro"))
+
+    return results
+
+# See https://towardsdatascience.com/fine-tune-your-llm-without-maxing-out-your-gpu-db2278603d78 for details
+trainer = transformers.Trainer(
+    model=lora_model,
+    train_dataset=model_params['train_data'],
+    eval_dataset=model_params['validation_data'],
+    compute_metrics=compute_metrics,
+    args=transformers.TrainingArguments(
+        per_device_train_batch_size=model_params['batch_size'],
+        per_device_eval_batch_size=model_params['batch_size'],
+        gradient_accumulation_steps=model_params['gradient_accumulation'],
+        warmup_steps=100,
+        max_steps=12276,
+        learning_rate=model_params['lr'],
+        fp16=True,
+        eval_steps= 1000,
+        logging_steps=1000,
+        save_steps=1000,
+        evaluation_strategy=model_params['evaluation_strategy'],
+        do_eval=True,
+        load_best_model_at_end=True,
+        metric_for_best_model="f1",
+        output_dir='model_outputs',
+        logging_dir='model_outputs',
+        remove_unused_columns =False, 
+        report_to='wandb'  # enable logging to W&B
+    ),
 )
-
 trainer.train()

fine_tune_data/validation_data.csv

@@ -0,0 +1,2 @@
+text
+"### Human: What is the aerospace mechanisms symposia?### Assistant: An annual meeting of space mechanism experts. {'source': 'DM', 'page': 0}"

playground.ipynb

@@ -1,64 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from transformers import AutoModel\n",
-    "import torch"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "def calculate_memory_required(model_name):\n",
-    "    model = AutoModel.from_pretrained(model_name)\n",
-    "\n",
-    "    # Calculate total parameters (assuming model parameters and gradients are in FP32)\n",
-    "    total_params = sum(p.numel() for p in model.parameters())\n",
-    "    total_memory_params = total_params * 4  # 4 bytes for FP32\n",
-    "\n",
-    "    # Optimizer states (e.g., for Adam, it's roughly the same as the model parameters)\n",
-    "    optimizer_memory = total_memory_params * 2  # Adam stores two values per parameter\n",
-    "\n",
-    "    # Batch size and sequence length\n",
-    "    batch_size = 32\n",
-    "    sequence_length = 512\n",
-    "    # Estimate activation memory (very rough estimate)\n",
-    "    activation_memory_per_example = sequence_length * model.config.hidden_size * 4  # 4 bytes for FP32\n",
-    "    total_activation_memory = batch_size * activation_memory_per_example\n",
-    "\n",
-    "    # Total estimated memory\n",
-    "    total_estimated_memory = total_memory_params + optimizer_memory + total_activation_memory\n",
-    "\n",
-    "    print(f\"Estimated memory for model and gradients: {total_memory_params / (1024 ** 3):.2f} GB\")\n",
-    "    print(f\"Estimated memory for optimizer states: {optimizer_memory / (1024 ** 3):.2f} GB\")\n",
-    "    print(f\"Estimated memory for activations: {total_activation_memory / (1024 ** 3):.2f} GB\")\n",
-    "    print(f\"Total estimated memory: {total_estimated_memory / (1024 ** 3):.2f} GB\")\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# Load model\n",
-    "model_name = 'mistralai/Mistral-7B-v0.1'\n",
-    "calculate_memory_required(model_name)\n"
-   ]
-  }
- ],
- "metadata": {
-  "language_info": {
-   "name": "python"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 2
-}

poetry.lock

@@ -186,6 +186,17 @@ docs = ["furo", "myst-parser", "sphinx", "sphinx-notfound-page", "sphinxcontrib-
 tests = ["attrs[tests-no-zope]", "zope-interface"]
 tests-no-zope = ["cloudpickle", "hypothesis", "mypy (>=1.1.1)", "pympler", "pytest (>=4.3.0)", "pytest-mypy-plugins", "pytest-xdist[psutil]"]
 
+[[package]]
+name = "bitsandbytes"
+version = "0.41.3.post2"
+description = "k-bit optimizers and matrix multiplication routines."
+optional = false
+python-versions = "*"
+files = [
+    {file = "bitsandbytes-0.41.3.post2-py3-none-any.whl", hash = "sha256:ceb301a3d4e6bf52bdad8d09f3064ac194bdfdeae535994c0315bd2ef7639cca"},
+    {file = "bitsandbytes-0.41.3.post2.tar.gz", hash = "sha256:7d25a51fb3b74b58e569473f8b70a5239124c0593dc053479c41cf2cd6730502"},
+]
+
 [[package]]
 name = "certifi"
 version = "2023.11.17"
@@ -495,6 +506,42 @@ files = [
     {file = "docstring_parser-0.15.tar.gz", hash = "sha256:48ddc093e8b1865899956fcc03b03e66bb7240c310fac5af81814580c55bf682"},
 ]
 
+[[package]]
+name = "evaluate"
+version = "0.4.1"
+description = "HuggingFace community-driven open-source library of evaluation"
+optional = false
+python-versions = ">=3.7.0"
+files = [
+    {file = "evaluate-0.4.1-py3-none-any.whl", hash = "sha256:3ff079ab09572c0a2c1e6d749887c19f6783ab993320412cd39f6fe501d28510"},
+    {file = "evaluate-0.4.1.tar.gz", hash = "sha256:d721d9f2059ced79770d8a0509e954fbd1bbac96a8f9160e29888d8073cda3d9"},
+]
+
+[package.dependencies]
+datasets = ">=2.0.0"
+dill = "*"
+fsspec = {version = ">=2021.05.0", extras = ["http"]}
+huggingface-hub = ">=0.7.0"
+multiprocess = "*"
+numpy = ">=1.17"
+packaging = "*"
+pandas = "*"
+requests = ">=2.19.0"
+responses = "<0.19"
+tqdm = ">=4.62.1"
+xxhash = "*"
+
+[package.extras]
+dev = ["Werkzeug (>=1.0.1)", "absl-py", "accelerate", "bert-score (>=0.3.6)", "black (>=22.0,<23.0)", "cer (>=1.2.0)", "charcut (>=1.1.1)", "flake8 (>=3.8.3)", "isort (>=5.0.0)", "jiwer", "mauve-text", "nltk", "pytest", "pytest-datadir", "pytest-xdist", "pyyaml (>=5.3.1)", "requests-file (>=1.5.1)", "rouge-score (>=0.1.2)", "sacrebleu", "sacremoses", "scikit-learn", "scipy", "sentencepiece", "seqeval", "six (>=1.15.0,<1.16.0)", "tensorflow (>=2.3,!=2.6.0,!=2.6.1,<=2.10)", "texttable (>=1.6.3)", "tldextract (>=3.1.0)", "toml (>=0.10.1)", "torch", "transformers", "trectools", "unidecode (>=1.3.4)"]
+docs = ["s3fs"]
+evaluator = ["scipy (>=1.7.1)", "transformers"]
+quality = ["black (>=22.0,<23.0)", "flake8 (>=3.8.3)", "isort (>=5.0.0)", "pyyaml (>=5.3.1)"]
+template = ["cookiecutter", "gradio (>=3.0.0)"]
+tensorflow = ["tensorflow (>=2.2.0,!=2.6.0,!=2.6.1)"]
+tensorflow-gpu = ["tensorflow-gpu (>=2.2.0,!=2.6.0,!=2.6.1)"]
+tests = ["Werkzeug (>=1.0.1)", "absl-py", "accelerate", "bert-score (>=0.3.6)", "cer (>=1.2.0)", "charcut (>=1.1.1)", "jiwer", "mauve-text", "nltk", "pytest", "pytest-datadir", "pytest-xdist", "requests-file (>=1.5.1)", "rouge-score (>=0.1.2)", "sacrebleu", "sacremoses", "scikit-learn", "scipy", "sentencepiece", "seqeval", "six (>=1.15.0,<1.16.0)", "tensorflow (>=2.3,!=2.6.0,!=2.6.1,<=2.10)", "texttable (>=1.6.3)", "tldextract (>=3.1.0)", "toml (>=0.10.1)", "torch", "transformers", "trectools", "unidecode (>=1.3.4)"]
+torch = ["torch"]
+
 [[package]]
 name = "executing"
 version = "2.0.1"
@@ -797,6 +844,17 @@ MarkupSafe = ">=2.0"
 [package.extras]
 i18n = ["Babel (>=2.7)"]
 
+[[package]]
+name = "joblib"
+version = "1.3.2"
+description = "Lightweight pipelining with Python functions"
+optional = false
+python-versions = ">=3.7"
+files = [
+    {file = "joblib-1.3.2-py3-none-any.whl", hash = "sha256:ef4331c65f239985f3f2220ecc87db222f08fd22097a3dd5698f693875f8cbb9"},
+    {file = "joblib-1.3.2.tar.gz", hash = "sha256:92f865e621e17784e7955080b6d042489e3b8e294949cc44c6eac304f59772b1"},
+]
+
 [[package]]
 name = "jupyter-client"
 version = "8.6.0"
@@ -1933,6 +1991,24 @@ urllib3 = ">=1.21.1,<3"
 socks = ["PySocks (>=1.5.6,!=1.5.7)"]
 use-chardet-on-py3 = ["chardet (>=3.0.2,<6)"]
 
+[[package]]
+name = "responses"
+version = "0.18.0"
+description = "A utility library for mocking out the `requests` Python library."
+optional = false
+python-versions = ">=3.7"
+files = [
+    {file = "responses-0.18.0-py3-none-any.whl", hash = "sha256:15c63ad16de13ee8e7182d99c9334f64fd81f1ee79f90748d527c28f7ca9dd51"},
+    {file = "responses-0.18.0.tar.gz", hash = "sha256:380cad4c1c1dc942e5e8a8eaae0b4d4edf708f4f010db8b7bcfafad1fcd254ff"},
+]
+
+[package.dependencies]
+requests = ">=2.0,<3.0"
+urllib3 = ">=1.25.10"
+
+[package.extras]
+tests = ["coverage (>=6.0.0)", "flake8", "mypy", "pytest (>=4.6)", "pytest-cov", "pytest-localserver", "types-mock", "types-requests"]
+
 [[package]]
 name = "rich"
 version = "13.7.0"
@@ -2070,6 +2146,95 @@ tensorflow = ["safetensors[numpy]", "tensorflow (>=2.11.0)"]
 testing = ["h5py (>=3.7.0)", "huggingface_hub (>=0.12.1)", "hypothesis (>=6.70.2)", "pytest (>=7.2.0)", "pytest-benchmark (>=4.0.0)", "safetensors[numpy]", "setuptools_rust (>=1.5.2)"]
 torch = ["safetensors[numpy]", "torch (>=1.10)"]
 
+[[package]]
+name = "scikit-learn"
+version = "1.3.2"
+description = "A set of python modules for machine learning and data mining"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "scikit-learn-1.3.2.tar.gz", hash = "sha256:a2f54c76accc15a34bfb9066e6c7a56c1e7235dda5762b990792330b52ccfb05"},
+    {file = "scikit_learn-1.3.2-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:e326c0eb5cf4d6ba40f93776a20e9a7a69524c4db0757e7ce24ba222471ee8a1"},
+    {file = "scikit_learn-1.3.2-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:535805c2a01ccb40ca4ab7d081d771aea67e535153e35a1fd99418fcedd1648a"},
+    {file = "scikit_learn-1.3.2-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1215e5e58e9880b554b01187b8c9390bf4dc4692eedeaf542d3273f4785e342c"},
+    {file = "scikit_learn-1.3.2-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:0ee107923a623b9f517754ea2f69ea3b62fc898a3641766cb7deb2f2ce450161"},
+    {file = "scikit_learn-1.3.2-cp310-cp310-win_amd64.whl", hash = "sha256:35a22e8015048c628ad099da9df5ab3004cdbf81edc75b396fd0cff8699ac58c"},
+    {file = "scikit_learn-1.3.2-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:6fb6bc98f234fda43163ddbe36df8bcde1d13ee176c6dc9b92bb7d3fc842eb66"},
+    {file = "scikit_learn-1.3.2-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:18424efee518a1cde7b0b53a422cde2f6625197de6af36da0b57ec502f126157"},
+    {file = "scikit_learn-1.3.2-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:3271552a5eb16f208a6f7f617b8cc6d1f137b52c8a1ef8edf547db0259b2c9fb"},
+    {file = "scikit_learn-1.3.2-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc4144a5004a676d5022b798d9e573b05139e77f271253a4703eed295bde0433"},
+    {file = "scikit_learn-1.3.2-cp311-cp311-win_amd64.whl", hash = "sha256:67f37d708f042a9b8d59551cf94d30431e01374e00dc2645fa186059c6c5d78b"},
+    {file = "scikit_learn-1.3.2-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:8db94cd8a2e038b37a80a04df8783e09caac77cbe052146432e67800e430c028"},
+    {file = "scikit_learn-1.3.2-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:61a6efd384258789aa89415a410dcdb39a50e19d3d8410bd29be365bcdd512d5"},
+    {file = "scikit_learn-1.3.2-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:cb06f8dce3f5ddc5dee1715a9b9f19f20d295bed8e3cd4fa51e1d050347de525"},
+    {file = "scikit_learn-1.3.2-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:5b2de18d86f630d68fe1f87af690d451388bb186480afc719e5f770590c2ef6c"},
+    {file = "scikit_learn-1.3.2-cp312-cp312-win_amd64.whl", hash = "sha256:0402638c9a7c219ee52c94cbebc8fcb5eb9fe9c773717965c1f4185588ad3107"},
+    {file = "scikit_learn-1.3.2-cp38-cp38-macosx_10_9_x86_64.whl", hash = "sha256:a19f90f95ba93c1a7f7924906d0576a84da7f3b2282ac3bfb7a08a32801add93"},
+    {file = "scikit_learn-1.3.2-cp38-cp38-macosx_12_0_arm64.whl", hash = "sha256:b8692e395a03a60cd927125eef3a8e3424d86dde9b2370d544f0ea35f78a8073"},
+    {file = "scikit_learn-1.3.2-cp38-cp38-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:15e1e94cc23d04d39da797ee34236ce2375ddea158b10bee3c343647d615581d"},
+    {file = "scikit_learn-1.3.2-cp38-cp38-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:785a2213086b7b1abf037aeadbbd6d67159feb3e30263434139c98425e3dcfcf"},
+    {file = "scikit_learn-1.3.2-cp38-cp38-win_amd64.whl", hash = "sha256:64381066f8aa63c2710e6b56edc9f0894cc7bf59bd71b8ce5613a4559b6145e0"},
+    {file = "scikit_learn-1.3.2-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:6c43290337f7a4b969d207e620658372ba3c1ffb611f8bc2b6f031dc5c6d1d03"},
+    {file = "scikit_learn-1.3.2-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:dc9002fc200bed597d5d34e90c752b74df516d592db162f756cc52836b38fe0e"},
+    {file = "scikit_learn-1.3.2-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:1d08ada33e955c54355d909b9c06a4789a729977f165b8bae6f225ff0a60ec4a"},
+    {file = "scikit_learn-1.3.2-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:763f0ae4b79b0ff9cca0bf3716bcc9915bdacff3cebea15ec79652d1cc4fa5c9"},
+    {file = "scikit_learn-1.3.2-cp39-cp39-win_amd64.whl", hash = "sha256:ed932ea780517b00dae7431e031faae6b49b20eb6950918eb83bd043237950e0"},
+]
+
+[package.dependencies]
+joblib = ">=1.1.1"
+numpy = ">=1.17.3,<2.0"
+scipy = ">=1.5.0"
+threadpoolctl = ">=2.0.0"
+
+[package.extras]
+benchmark = ["matplotlib (>=3.1.3)", "memory-profiler (>=0.57.0)", "pandas (>=1.0.5)"]
+docs = ["Pillow (>=7.1.2)", "matplotlib (>=3.1.3)", "memory-profiler (>=0.57.0)", "numpydoc (>=1.2.0)", "pandas (>=1.0.5)", "plotly (>=5.14.0)", "pooch (>=1.6.0)", "scikit-image (>=0.16.2)", "seaborn (>=0.9.0)", "sphinx (>=6.0.0)", "sphinx-copybutton (>=0.5.2)", "sphinx-gallery (>=0.10.1)", "sphinx-prompt (>=1.3.0)", "sphinxext-opengraph (>=0.4.2)"]
+examples = ["matplotlib (>=3.1.3)", "pandas (>=1.0.5)", "plotly (>=5.14.0)", "pooch (>=1.6.0)", "scikit-image (>=0.16.2)", "seaborn (>=0.9.0)"]
+tests = ["black (>=23.3.0)", "matplotlib (>=3.1.3)", "mypy (>=1.3)", "numpydoc (>=1.2.0)", "pandas (>=1.0.5)", "pooch (>=1.6.0)", "pyamg (>=4.0.0)", "pytest (>=7.1.2)", "pytest-cov (>=2.9.0)", "ruff (>=0.0.272)", "scikit-image (>=0.16.2)"]
+
+[[package]]
+name = "scipy"
+version = "1.11.4"
+description = "Fundamental algorithms for scientific computing in Python"
+optional = false
+python-versions = ">=3.9"
+files = [
+    {file = "scipy-1.11.4-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:bc9a714581f561af0848e6b69947fda0614915f072dfd14142ed1bfe1b806710"},
+    {file = "scipy-1.11.4-cp310-cp310-macosx_12_0_arm64.whl", hash = "sha256:cf00bd2b1b0211888d4dc75656c0412213a8b25e80d73898083f402b50f47e41"},
+    {file = "scipy-1.11.4-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b9999c008ccf00e8fbcce1236f85ade5c569d13144f77a1946bef8863e8f6eb4"},
+    {file = "scipy-1.11.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:933baf588daa8dc9a92c20a0be32f56d43faf3d1a60ab11b3f08c356430f6e56"},
+    {file = "scipy-1.11.4-cp310-cp310-musllinux_1_1_x86_64.whl", hash = "sha256:8fce70f39076a5aa62e92e69a7f62349f9574d8405c0a5de6ed3ef72de07f446"},
+    {file = "scipy-1.11.4-cp310-cp310-win_amd64.whl", hash = "sha256:6550466fbeec7453d7465e74d4f4b19f905642c89a7525571ee91dd7adabb5a3"},
+    {file = "scipy-1.11.4-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:f313b39a7e94f296025e3cffc2c567618174c0b1dde173960cf23808f9fae4be"},
+    {file = "scipy-1.11.4-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:1b7c3dca977f30a739e0409fb001056484661cb2541a01aba0bb0029f7b68db8"},
+    {file = "scipy-1.11.4-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:00150c5eae7b610c32589dda259eacc7c4f1665aedf25d921907f4d08a951b1c"},
+    {file = "scipy-1.11.4-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:530f9ad26440e85766509dbf78edcfe13ffd0ab7fec2560ee5c36ff74d6269ff"},
+    {file = "scipy-1.11.4-cp311-cp311-musllinux_1_1_x86_64.whl", hash = "sha256:5e347b14fe01003d3b78e196e84bd3f48ffe4c8a7b8a1afbcb8f5505cb710993"},
+    {file = "scipy-1.11.4-cp311-cp311-win_amd64.whl", hash = "sha256:acf8ed278cc03f5aff035e69cb511741e0418681d25fbbb86ca65429c4f4d9cd"},
+    {file = "scipy-1.11.4-cp312-cp312-macosx_10_9_x86_64.whl", hash = "sha256:028eccd22e654b3ea01ee63705681ee79933652b2d8f873e7949898dda6d11b6"},
+    {file = "scipy-1.11.4-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:2c6ff6ef9cc27f9b3db93a6f8b38f97387e6e0591600369a297a50a8e96e835d"},
+    {file = "scipy-1.11.4-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:b030c6674b9230d37c5c60ab456e2cf12f6784596d15ce8da9365e70896effc4"},
+    {file = "scipy-1.11.4-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:ad669df80528aeca5f557712102538f4f37e503f0c5b9541655016dd0932ca79"},
+    {file = "scipy-1.11.4-cp312-cp312-musllinux_1_1_x86_64.whl", hash = "sha256:ce7fff2e23ab2cc81ff452a9444c215c28e6305f396b2ba88343a567feec9660"},
+    {file = "scipy-1.11.4-cp312-cp312-win_amd64.whl", hash = "sha256:36750b7733d960d7994888f0d148d31ea3017ac15eef664194b4ef68d36a4a97"},
+    {file = "scipy-1.11.4-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:6e619aba2df228a9b34718efb023966da781e89dd3d21637b27f2e54db0410d7"},
+    {file = "scipy-1.11.4-cp39-cp39-macosx_12_0_arm64.whl", hash = "sha256:f3cd9e7b3c2c1ec26364856f9fbe78695fe631150f94cd1c22228456404cf1ec"},
+    {file = "scipy-1.11.4-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:d10e45a6c50211fe256da61a11c34927c68f277e03138777bdebedd933712fea"},
+    {file = "scipy-1.11.4-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:91af76a68eeae0064887a48e25c4e616fa519fa0d38602eda7e0f97d65d57937"},
+    {file = "scipy-1.11.4-cp39-cp39-musllinux_1_1_x86_64.whl", hash = "sha256:6df1468153a31cf55ed5ed39647279beb9cfb5d3f84369453b49e4b8502394fd"},
+    {file = "scipy-1.11.4-cp39-cp39-win_amd64.whl", hash = "sha256:ee410e6de8f88fd5cf6eadd73c135020bfbbbdfcd0f6162c36a7638a1ea8cc65"},
+    {file = "scipy-1.11.4.tar.gz", hash = "sha256:90a2b78e7f5733b9de748f589f09225013685f9b218275257f8a8168ededaeaa"},
+]
+
+[package.dependencies]
+numpy = ">=1.21.6,<1.28.0"
+
+[package.extras]
+dev = ["click", "cython-lint (>=0.12.2)", "doit (>=0.36.0)", "mypy", "pycodestyle", "pydevtool", "rich-click", "ruff", "types-psutil", "typing_extensions"]
+doc = ["jupytext", "matplotlib (>2)", "myst-nb", "numpydoc", "pooch", "pydata-sphinx-theme (==0.9.0)", "sphinx (!=4.1.0)", "sphinx-design (>=0.2.0)"]
+test = ["asv", "gmpy2", "mpmath", "pooch", "pytest", "pytest-cov", "pytest-timeout", "pytest-xdist", "scikit-umfpack", "threadpoolctl"]
+
 [[package]]
 name = "shtab"
 version = "1.6.5"
@@ -2128,6 +2293,17 @@ files = [
 [package.dependencies]
 mpmath = ">=0.19"
 
+[[package]]
+name = "threadpoolctl"
+version = "3.2.0"
+description = "threadpoolctl"
+optional = false
+python-versions = ">=3.8"
+files = [
+    {file = "threadpoolctl-3.2.0-py3-none-any.whl", hash = "sha256:2b7818516e423bdaebb97c723f86a7c6b0a83d3f3b0970328d66f4d9104dc032"},
+    {file = "threadpoolctl-3.2.0.tar.gz", hash = "sha256:c96a0ba3bdddeaca37dc4cc7344aafad41cdb8c313f74fdfe387a867bba93355"},
+]
+
 [[package]]
 name = "tokenizers"
 version = "0.15.0"
@@ -2765,4 +2941,4 @@ multidict = ">=4.0"
 [metadata]
 lock-version = "2.0"
 python-versions = "^3.11"
-content-hash = "bcc7e7ed0cdbb6526fc703d12aa9069073276b06eb072bf2f1edf4645d9492a2"
+content-hash = "911f7cb8678df6c8f7f0151945feedf5d54d8fcefdcc2339e9eb90360b82c97a"

pyproject.toml

@@ -13,6 +13,9 @@ transformers = "^4.36.2"
 torch = "^2.1.2"
 ipykernel = "^6.27.1"
 peft = "^0.7.1"
+bitsandbytes = "^0.41.3.post2"
+evaluate = "^0.4.1"
+scikit-learn = "^1.3.2"
 
 
 [build-system]

requirements.txt

@@ -10,6 +10,7 @@ async-lru==2.0.4
 attrs==23.1.0
 Babel==2.14.0
 beautifulsoup4==4.12.2
+bitsandbytes==0.41.3.post2
 bleach==6.1.0
 certifi==2023.11.17
 cffi==1.16.0
@@ -21,6 +22,7 @@ decorator==5.1.1
 defusedxml==0.7.1
 dill==0.3.7
 docstring-parser==0.15
+evaluate==0.4.1
 executing==2.0.1
 fastjsonschema==2.19.0
 filelock==3.13.1
@@ -35,6 +37,7 @@ ipywidgets==8.1.1
 isoduration==20.11.0
 jedi==0.19.1
 Jinja2==3.1.2
+joblib==1.3.2
 json5==0.9.14
 jsonpointer==2.4
 jsonschema==4.20.0
@@ -106,11 +109,14 @@ QtPy==2.4.1
 referencing==0.32.0
 regex==2023.10.3
 requests==2.31.0
+responses==0.18.0
 rfc3339-validator==0.1.4
 rfc3986-validator==0.1.1
 rich==13.7.0
 rpds-py==0.15.2
 safetensors==0.4.1
+scikit-learn==1.3.2
+scipy==1.11.4
 Send2Trash==1.8.2
 shtab==1.6.5
 six==1.16.0
@@ -119,6 +125,7 @@ soupsieve==2.5
 stack-data==0.6.3
 sympy==1.12
 terminado==0.18.0
+threadpoolctl==3.2.0
 tinycss2==1.2.1
 tokenizers==0.15.0
 torch==2.1.2